1299412 - Crash in dalvik-main space 1 (deleted)@0xfb56df9

Status: RESOLVED WONTFIX

(Firefox for Android Graveyard :: General, defect)

Description

[Pierre Ossman]

This bug was filed from the Socorro interface and is 
report bp-8e31a13f-768d-4b1b-8e6b-b0cf02160831.
=============================================================

Fully reproducible on Samsung S6 using Android 6.0.1. Also happens on Nexus 7 with Android 5.1.1:

https://crash-stats.mozilla.com/report/index/45c328ea-4982-4e19-9dc0-6858a2160831

Does not happen with Firefox 46 on either the Samsung S6 or the Nexus 7. I.e. this is a regression.

(One Samsung S6 also fails to report the crashes for some reason when this happens. See bug 1288892).

The page provoking this is currently not public. It is a modified version of noVNC with a bunch of shadows and opacity. Seems to happen when alpha regions are above a canvas that is being modified. Trying to pinpoint better what provokes the crash.

Comment 1

[Pierre Ossman]

Hmm... take the Nexus 7 report with a grain of salt. It's a different crash and we're having trouble reproducing the crash on that device. The Samsung S6 explodes reliably though.

Comment 2

[Ludovic Hirlimann [:Usul]]

> The page provoking this is currently not public. It is a modified version of
> noVNC with a bunch of shadows and opacity. Seems to happen when alpha
> regions are above a canvas that is being modified. Trying to pinpoint better
> what provokes the crash.

can you try to write a mimimal testcase and attach it to the bug ?

Comment 3

[Pierre Ossman]

Maybe. If we can figure out what on the page is provoking the bug. :)

Comment 4

[Pierre Ossman]

The version of noVNC that provokes this bug is now publicly available:

https://github.com/kanaka/noVNC/pull/653

I'll try to find some time to make a minimal test case.

Comment 5

[Pierre Ossman]

I'm afraid I'm failing at making a smaller test case for this. I removed the main logic from the code and just put a simple animation on the canvas instead. Unfortunately that was enough to avoid the bug. So it seems like this requires a bit of a perfect storm to happen. :/

Comment 6

[James Willcox (:snorp) ([e-mail]) (he/him)]

Can you attach logcat from when the crash occurs?

Comment 7

[Pierre Ossman]

Attachment: logcat

Here's a logcat from the crash. Interesting part is:

> 09-08 13:44:34.242 14343 14343 W art     : Throwing OutOfMemoryError "Could not allocate JNI Env"
> 09-08 13:44:34.252 14343 14343 D AndroidRuntime: Shutting down VM
> 09-08 13:44:34.252  3105  4215 E BufferQueueProducer: [SurfaceView] dequeueBuffer: can't dequeue multiple buffers without setting the buffer count
> 09-08 13:44:34.252 14343 14343 E GeckoCrashHandler: >>> REPORTING UNCAUGHT EXCEPTION FROM THREAD 1 ("main")
> 09-08 13:44:34.252 14343 14343 E GeckoCrashHandler: java.lang.OutOfMemoryError: Could not allocate JNI Env

Comment 8

[Pierre Ossman]

Any clues from the log? Any more tracing I can do? Any clues as to what provokes it so I can make a smaller test case?

Comment 9

[Sylvestre Ledru [:Sylvestre]]

Closing because no crash reported since 12 weeks.

Comment 10

[Sylvestre Ledru [:Sylvestre]]

Closing because no crash reported since 12 weeks.